ENH BUG DOC TST fixed behavior of str_cat w.r.t. nan handling and added friendly error message

DOC corrected docstring for str_cat

closes #12297
closes #11435
closes #11334

Author: hack-c

doc/source/whatsnew/v0.18.0.txt

@@ -249,6 +249,24 @@ the ``extractall`` method returns all matches.
 
    s.str.extractall("(?P<letter>[ab])(?P<digit>\d)")
 
+Changes to str.cat
+^^^^^^^^^^^^^^^^^^
+
+The :ref:`.str.cat <text.cat>` concatenates the members of a ``Series``. Before, if ``NaN`` values were present in the Series, calling ``.str.cat()`` on it would return ``NaN``, unlike the rest of the ``Series.str.*`` API. This behavior has been amended to ignore ``NaN`` values by default. (:issue:`11435`).
+
+A new, friendlier ``ValueError`` is added to protect against the mistake of supplying the ``sep`` as an arg, rather than as a kwarg. (:issue:`11334`).
+
+.. ipython:: python
+
+    Series(['a','b',np.nan,'c']).str.cat(sep=' ')
+    Series(['a','b',np.nan,'c']).str.cat(sep=' ', na_rep='?')
+
+.. code-block:: python
+
+    In [2]: Series(['a','b',np.nan,'c']).str.cat(' ')
+    ValueError: Did you mean to supply a `sep` keyword?
+
+
 .. _whatsnew_0180.enhancements.rounding:
 
 Datetimelike rounding

pandas/core/strings.py

@@ -1,7 +1,8 @@
 import numpy as np
 
 from pandas.compat import zip
-from pandas.core.common import (isnull, _values_from_object, is_bool_dtype,
+from pandas.core.common import (isnull, notnull, _values_from_object,
+                                is_bool_dtype,
                                 is_list_like, is_categorical_dtype,
                                 is_object_dtype, take_1d)
 import pandas.compat as compat
@@ -37,14 +38,23 @@ def str_cat(arr, others=None, sep=None, na_rep=None):
       If None, returns str concatenating strings of the Series
     sep : string or None, default None
     na_rep : string or None, default None
-        If None, an NA in any array will propagate
+        If None, NA in the series are ignored.
 
     Returns
     -------
     concat : Series/Index of objects or str
 
     Examples
     --------
+    When ``na_rep`` is `None` (default behavior), NaN value(s)
+    in the Series are ignored.
+
+    >>> Series(['a','b',np.nan,'c']).str.cat(sep=' ')
+    'a b c'
+
+    >>> Series(['a','b',np.nan,'c']).str.cat(sep=' ', na_rep='?')
+    'a b ? c'
+
     If ``others`` is specified, corresponding values are
     concatenated with the separator. Result will be a Series of strings.
 
@@ -103,18 +113,23 @@ def str_cat(arr, others=None, sep=None, na_rep=None):
         arr = np.asarray(arr, dtype=object)
         mask = isnull(arr)
         if na_rep is None and mask.any():
-            return np.nan
+            if sep == '':
+                na_rep = ''
+            else:
+                return sep.join(arr[notnull(arr)])
         return sep.join(np.where(mask, na_rep, arr))
 
 
 def _length_check(others):
     n = None
     for x in others:
-        if n is None:
-            n = len(x)
-        elif len(x) != n:
-            raise ValueError('All arrays must be same length')
-
+        try:
+            if n is None:
+                n = len(x)
+            elif len(x) != n:
+                raise ValueError('All arrays must be same length')
+        except TypeError:
+            raise ValueError("Did you mean to supply a `sep` keyword?")
     return n
 
 

pandas/tests/test_strings.py

@@ -100,7 +100,8 @@ def test_cat(self):
 
         # single array
         result = strings.str_cat(one)
-        self.assertTrue(isnull(result))
+        exp = 'aabbc'
+        self.assertEqual(result, exp)
 
         result = strings.str_cat(one, na_rep='NA')
         exp = 'aabbcNA'
@@ -114,6 +115,10 @@ def test_cat(self):
         exp = 'a_a_b_b_c_NA'
         self.assertEqual(result, exp)
 
+        result = strings.str_cat(two, sep='-')
+        exp = 'a-b-d-foo'
+        self.assertEqual(result, exp)
+
         # Multiple arrays
         result = strings.str_cat(one, [two], na_rep='NA')
         exp = ['aa', 'aNA', 'bb', 'bd', 'cfoo', 'NANA']
@@ -2453,6 +2458,15 @@ def test_cat_on_filtered_index(self):
 
         self.assertEqual(str_multiple.loc[1], '2011 2 2')
 
+    def test_str_cat_raises_intuitive_error(self):
+        # https://github.com/pydata/pandas/issues/11334
+        s = Series(['a', 'b', 'c', 'd'])
+        message = "Did you mean to supply a `sep` keyword?"
+        with tm.assertRaisesRegexp(ValueError, message):
+            s.str.cat('|')
+        with tm.assertRaisesRegexp(ValueError, message):
+            s.str.cat('    ')
+
     def test_index_str_accessor_visibility(self):
         from pandas.core.strings import StringMethods